Jorge Muñoz, Developer in Madrid, Spain
Jorge is available for hire
Hire Jorge

Jorge Muñoz

Verified Expert  in Engineering

Artificial Intelligence (AI) Developer

Madrid, Spain
Toptal Member Since
February 27, 2020

Jorge is a deep learning researcher and engineer with a PhD in computer science and artificial intelligence and an MBA. He's successfully worked with startups based in San Francisco, London, and Madrid. Along with an expertise in AI, Jorge has several years of experience in the industry, including founding and running his own tech company. Since 2015, he’s been focusing primarily on remote roles.

Deep Reinforcement Learning, Artificial Intelligence (AI), Applied Research...
Realm Living Inc
Machine Learning, Image Processing, Neural Networks, Deep Learning...
ENGworks Inc.
Machine Learning, Building Information Modeling (BIM)...




Preferred Environment

Kotlin, Python, Deep Learning, IntelliJ IDEA, Neural Networks

The most amazing...

...project I've participated in involved conducting research about machine consciousness that was discussed in Science Magazine.

Work Experience

Reinforcement Learning Expert

2022 - 2022
  • Trained a MaskPPO to control 3D software to help users generate virtual scenarios.
  • Used hugging face NLP models to predict the following action to be executed based on a sequence of actions and states (reinforcement learning transformer).
  • Helped design how to collect data and train a model for a 3D design tool.
Technologies: Deep Reinforcement Learning, Artificial Intelligence (AI), Applied Research, Algorithms, Data Scientist

Machine Learning Expert

2021 - 2022
Realm Living Inc
  • Trained a model for predicting house rooms and quantitative objects inside them.
  • Created a methodology to train and deploy new models in the company, speeding up the time to prepare and deploy a new model to production to a couple of days.
  • Managed a junior developer to become as productive as a senior developer.
Technologies: Machine Learning, Image Processing, Neural Networks, Deep Learning, You Only Look Once (YOLO), Object Detection, Convolutional Neural Networks (CNN), EfficientNet, CoAtNet, Vision Transformer (ViT), Residual Neural Network (ResNet), Data Scientist, Leadership

Machine Learning Expert

2021 - 2021
ENGworks Inc.
  • Researched different models for instance semantic segmentation of objects in 3D point clouds.
  • Helped the client to create a dataset of 3D point clouds based on the requirements to train models, which would lead to quality predictions.
  • Improved models to work with 3D point clouds that were hundreds of times bigger than the original models.
Technologies: Machine Learning, Building Information Modeling (BIM), Machine Learning Automation, Models, Artificial Intelligence (AI), Data Scientist


2020 - 2020
Ognito LLC
  • Integrated different deep learning models to replace a picture's face with a fake (GAN-generated) face.
  • Modified the GAN work to make it create a face with the same shades and lights of a specific face.
  • Deployed the different models in a pipeline with different Docker containers.
Technologies: Computer Vision, Python, TensorFlow, Generative Adversarial Networks (GANs), PyTorch, Keras, Deep Reinforcement Learning, TensorFlow Deep Learning Library (TFLearn), Algorithms, Research

Consultant, Trading Predictions

2020 - 2020
NeuralX Limited
  • Researched different methods to apply deep learning to trading objectives.
  • Trained and tested different sequence models (transformers) for trading.
  • Researched reinforcement learning methods and how to apply them to trading.
Technologies: Artificial Intelligence (AI), Python, TensorFlow, Data Science, Artificial Neural Networks (ANN), Neural Networks, Pandas, Mathematics, Data Scientist

Founder | CEO | CTO

2018 - 2020
Serendeepia Research
  • Managed successful projects for Fortune 500 companies and led the development of B2C products with the latest advancements in deep learning.
  • Created a remote-first culture with a well-defined methodology to manage the projects: scrum and kanban boards.
  • Grew the company from three to seven people in a year and led the team.
Technologies: Reinforcement Learning, Generative Pre-trained Transformers (GPT), Natural Language Processing (NLP), Linux, Computer Vision, Variational Autoencoders, Python, Kotlin, Convolutional Neural Networks (CNN), Deep Reinforcement Learning, Generative Adversarial Networks (GANs), Natural Language Understanding (NLU), Artificial Intelligence (AI), Machine Learning, Jenkins, Kubernetes, Scrum, TensorFlow, Deep Learning, Algorithms, Data Scientist, Research, Leadership, Communication

Machine Learning Researcher (Remote Contractor)

2017 - 2018
Good AI Lab
  • Increased awareness of the company by creating tutorial and demos with TensorFlow that ran in the platform.
  • Composed blog posts that were featured on several publications in Medium with thousands of views.
  • Created a NLP model able to classify scientific papers in biotechnology.
Technologies: Generative Pre-trained Transformers (GPT), Natural Language Processing (NLP), Amazon Web Services (AWS), Neural Networks, Python, Natural Language Understanding (NLU), Artificial Intelligence (AI), Machine Learning, Deep Learning, TensorFlow, Data Scientist, Research

Remote Machine Learning Engineer/Researcher

2016 - 2018
  • Migrated a simple single-label image-classification model in Caffe to a multi-label model in TensorFlow with TensorFlow Serving. The new model had more labels (more than 100) and better precision and recall for every class (more than 90% accuracy).
  • Established a scientific procedure to measure the quality of the models.
  • Increased the competitive advantage of our product with an aesthetic model that was able to score the beautifulness of the images with a state of the art CNN model.
  • Optimized the competitive advantage of our product with a model that creates a perceptual hash of images using their local and semantic information which was used to find similar images.
Technologies: Amazon Web Services (AWS), Computer Vision, Neural Networks, Natural Language Understanding (NLU), Machine Learning, Artificial Intelligence (AI), Deep Learning, Convolutional Neural Networks (CNN), Python, TensorFlow, Data Scientist, Research

Android Lead Developer

2013 - 2017
Several Companies (Quipper, Shopcade, and Appgree)
  • Developed three apps from scratch with more than a million downloads using the latest Android tools.
  • Trained teams to keep on working on the apps and continued on improving them.
  • Created a recommender engine for an app's back end.
Technologies: Java, Kotlin, Android

Founder | CEO

2010 - 2013
  • Conducted research that was published in an article in Science Magazine.
  • Won international competitions with our new technology based on machine consciousness.
  • Created a simple 3D game engine for Android which also integrated a physics engine.
Technologies: Neural Networks, Machine Learning, Artificial Intelligence (AI), C++, Java, Video Games, Android, Marketing, Research, Leadership, Communication


2007 - 2011
  • Published papers in top conferences over the world.
  • Won international competitions with ideas based on my research.
  • Published several book chapters based on my ideas in machine consciousness and imitation learning.
Technologies: Linux, Machine Learning, Artificial Intelligence (AI), C++, Neural Networks, Data Scientist, Research, Communication

Diffusion Model for Mobile Devices
A custom implementation of diffusion models to generate images. The model is small enough to be ported to mobile devices to generate images. It uses a CLIP model to encode the text for the diffusion model.

LLM Transformer for Programming Languages
A custom implementation of a GPT-type model trained in a programming language to help developers. It can be trained in more than one programming language, and the model is small enough to run on a single GPU.

An implementation of Hash Nerf (neural radiance fields) in PyTorch to run in a local machine with a 4090 GPU.

NeRF is a method for synthesizing novel views of complex scenes by optimizing an underlying continuous volumetric scene function using a sparse set of input views. This implementation uses hash tables to speed up the process.

Face Replacement by Fake Face in Images with GANs (2020)

I create a prototype of a tool that was able to replace all the faces in an image by fake faces generated with a GAN model. I used StyleGAN to transfer the style of the original faces to the fake faces generated; this way, the lighting, shadows, and other visual elements matched with the original image.

Stock Trading Prediction (2020)

I tested different models to predict of stock prices differences. Among other things, I used wavelets to reduce the noise in the input data and transformers as well as sequence to sequence models to make the predictions.

Selfie Improver with GANs (2019)

I created a GAN architecture based on StarGAN and progressive growth of GANs in order to modify pictures of faces of people. The model was able to create images with good resolution and to modify some attributes of the faces the same way StarGAN did.

Toxic Speech Detector for Comments in the Internet (2019)

I helped to create and train a deep learning model that classifies comments in different categories in order to detect toxic comments, harming behaviors, threats, insults or identity hate. The deep learning model was a BiLSTM with 93% accuracy.

Image Retrieval Based on Semantic Information and an Aesthetic Score (2017-2018)

I trained two deep learning models that added some metadata to the images to perform image retrieval efficiently. We processed more than 3 million images a day and we needed our users to find the best images that fit their needs.

I trained two convolutional deep learning models—one to create a semantic hash of the images based on their content, and another one to asses the aesthetic of the image—so we can rank the best matching images based on the engagement they could have.

Document Classification with a Hierarchical Attention Network (2017)
I created and trained a deep learning model for natural language processing using a hierarchical attention network. The goal was to classify scientific papers about cancer treatment to help in the cancer research. Among other things, I tested classic algorithms for text classification as bag of words and TF-IDF. For the deep learning models I used GRU and LSTM cells and an architecture with hierarchical attention. More complex models as BERT or GTP-2 were not available at that time.
2020 - 2024

Progress Toward Bachelor's Degree in Mathematics

UNED - Spain

2011 - 2012

Master of Business Administration (MBA) Degree in Business Administration

EOI | Escuela de Organización Industrial - Madrid, Spain

2008 - 2011

Ph.D. in Computer Science and Artificial Intelligence

Universidad Carlos III of Madrid - Madrid, Spain

2007 - 2008

Master's Degree in Computer Science and Information Technology

Universidad Carlos III of Madrid - Madrid, Spain

2001 - 2006

Bachelor's Degree in Computer Science Engineering

Universidad Carlos III of Madrid - Madrid, Spain


TensorFlow, Keras, PyTorch, Pandas, TensorFlow Deep Learning Library (TFLearn)


Jenkins, IntelliJ IDEA, You Only Look Once (YOLO)


Python, Kotlin, Java, C++


Linux, Android, Kubernetes, Amazon Web Services (AWS)


Data Science, Scrum, Building Information Modeling (BIM), Management

Industry Expertise





Artificial Intelligence (AI), Applied Research, Convolutional Neural Networks (CNN), Machine Learning, Deep Learning, Neural Networks, Deep Neural Networks, Algorithms, Data Scientist, Research, Computer Vision, Predictive Analytics, Leadership, Communication, Generative Pre-trained Transformer 3 (GPT-3), Large Language Models (LLMs), APIs, Applied Mathematics, Natural Language Understanding (NLU), Natural Language Processing (NLP), Deep Reinforcement Learning, Video Games, Reinforcement Learning, Variational Autoencoders, Generative Adversarial Networks (GANs), Artificial Neural Networks (ANN), Mathematics, Machine Learning Automation, Models, Image Processing, Object Detection, EfficientNet, CoAtNet, Vision Transformer (ViT), Residual Neural Network (ResNet), Generative Pre-trained Transformers (GPT), Language Models, Machine Learning Operations (MLOps), Diffusion Models

Collaboration That Works

How to Work with Toptal

Toptal matches you directly with global industry experts from our network in hours—not weeks or months.


Share your needs

Discuss your requirements and refine your scope in a call with a Toptal domain expert.

Choose your talent

Get a short list of expertly matched talent within 24 hours to review, interview, and choose from.

Start your risk-free talent trial

Work with your chosen talent on a trial basis for up to two weeks. Pay only if you decide to hire them.

Top talent is in high demand.

Start hiring